6068
J. Agric. Food Chem. 2001, 49, 6068−6078
Determination of Total N-Nitroso Compounds and Their Precursors in Frankfurters, Fresh Meat, Dried Salted Fish, Sauces, Tobacco, and Tobacco Smoke Particulates James Haorah,† Lin Zhou,† Xiaojie Wang,† Guoping Xu,† and Sidney S. Mirvish*,†,‡ Eppley Institute for Research in Cancer, Department of Pharmaceutical Sciences, and Department of Biochemistry and Molecular Biology, University of Nebraska Medical Center, Omaha, Nebraska 68198
Total N-nitroso compounds (NOC) and NOC precursors (NOCP) were determined in extracts of food and tobacco products. Following Walters’ method, NOC were decomposed to NO with refluxing HBr/HCl/HOAc/EtOAc and NO was measured by chemiluminescence. NOC were determined after sulfamic acid treatment to destroy nitrite, and NOCP were determined after treatment with 110 mM nitrite and then sulfamic acid. Analysis without HBr gave results e20% of those with HBr. This NOC method was efficient for nitrosamines but not nitrosoureas. The standard nitrosation for determining NOCP gave high yields for readily nitrosated amines, including 1-deoxy-1-fructosylvaline, but not for simple amines, dipeptides, and alkylureas. Mean NOC and NOCP results were (respectively, in µmol/kg of product) 5.5 and 2700 for frankfurters, 0.5 and 660 for fresh meat, 5.8 and 5800 for salted, dried fish, and 660 and 2900 for chewing tobacco (all for aqueous extracts) and 220 and 20000 nmol/cigarette for MeCN extracts of cigarette smoke filter pads. Keywords: N-Nitroso compounds; nitrosation; thermal energy analysis; frankfurters; meat; salted, dried fish; sauces; tobacco; cigarette smoke INTRODUCTION
HBr
N-Nitroso compounds (NOC) comprise nitrosamines (R1R2NNO) and nitrosamides [R1CON(NO)R2], produced by the nitrosation of secondary amines and N-alkyl amides, respectively (1). Most NOC induce cancer in laboratory animals (2, 3), and NOC may be involved in the etiology of several human cancers (4, 5). People are exposed to NOC in foods, by smoking or chewing tobacco products, and by the in vivo formation of NOC, for example, in the stomach from the acid-catalyzed reaction with nitrite of dietary amines and amides that are NOC precursors (NOCP) (4). NOCP may be converted to NOC during food storage or cooking or in vivo. Total NOC in human gastric juice [mean level ) 2.5 µmol/L (6)] and feces [mean level for subjects on a high-meat diet ) 23 µmol/kg (7)] could arise from dietary NOC and from the in vivo nitrosation of NOCP. NOC can be determined by Walters’ method, involving treatment of NOC with HBr to yield NO. The NO is determined by thermal energy analysis (TEA, chemiluminescence), in which the NO reacts with ozone to give NO2 in an excited state, which emits infrared light. This is measured with a detection limit of 100 pmol. The method is based on the finding (8) that HBr, but not HCl, reacts with NOC to produce nitrite, presumably via the reaction shown below. TEA can be used to determine total NOC by the Walters method or can be linked to gas chromatography (9) or HPLC (10) to determine individual NOC. Other methods for deter* Author to whom correspondence should be addressed [telephone (402) 559-5272; fax (402) 559-4651; e-mail smirvish@ unmc.edu). † Eppley Institute for Research in Cancer. ‡ Department of Pharmaceutical Sciences and Department of Biochemistry and Molecular Biology.
R1R2NNO 98 R1R2NH + NOBr f [O]
Br2 + NO 98 NO2mining individual NOC include gas chromatographymass spectrometry (11). The Walters method was described in 1976-1978 (12, 13), was further developed (14-19), and has been reviewed (15, 20, 21). This method was used to determine NOC in nitrite-cured meat (20, 22, 23) and in human gastric juice (17, 19), urine (18), and feces (7, 24). In those materials studied in detail, 70% of these NOC have not been identified is of concern. Therefore, we have begun to purify and identify the NOCP (which are far more abundant than the NOC) in frankfurters (28) and wished to determine the levels at which NOC and NOCP occur in various foods. We define NOCP as compounds that yield NOC after nitrosation under our standard conditions. NOCP represent the potential for NOC formation under these relatively mild conditions. The NOC method used here was modified from two studies on the analysis of human gastric juice. Pignatelli et al. (17) treated gastric juice with sulfamic acid (SA) and then analyzed the product by two TEA systems, one using HCl in an HOAc/EtOAc mixture (the “HCl mode”) and the other using HBr in a similar mixture (the “HBr mode”). TEA-responsive compounds included (a) inorganic nitrite, which was determined in both the HCl and HBr modes and was destroyed by SA (17); (b) unidentified “TEA-responsive compounds” that were not decomposed by SA and produced NO in the HCl and (it was presumed) the HBr mode; and (c) NOC, which were also not destroyed by SA and produced NO in the HBr but not the HCl mode (8). The difference between results
10.1021/jf010602h CCC: $20.00 © 2001 American Chemical Society Published on Web 11/16/2001
N-Nitroso Compounds and Their Precurors
obtained by the HBr and HCl modes after treatment with SA was taken to give the NOC level. On this basis, about half the TEA response by the HBr mode for SAtreated human gastric juice was attributed to NOC and about half to other TEA-responsive compounds. Xu and Reed (19) also defined NOC in gastric juice as the difference between results obtained by the HBr and HCl modes, but they inserted additional traps downstream from the reaction flask and included both HCl and HBr in analyses by the HBr mode. Sodium nitrite is added to meat and fish products to preserve them and to improve their taste and color (29). The hypothesis that these products were involved in the etiology of stomach cancer was introduced (30, 31) to help explain the high incidence of stomach cancer (a) in developing countries and (b) in the United States and Europe before the 1940s, when stomach cancer was the principal type of cancer (32). In both of these situations, meat and fish are/were often preserved with excessive amounts of salt [a promoting factor for stomach cancer (31)] and nitrite and nitrate relative to the controlled conditions of modern manufacture. Processed meat (mainly sausages) was significantly linked with colon cancer in two of four prospective studies reported in 1990-1994 (33-35) and in a current prospective study on nearly 500000 subjects from nine European countries (preliminary report by E. Riboli, Conference on Diet and Nutrition, American Institute for Cancer Research, 2001). Most of nine case-control studies reported associations between the consumption of frankfurters (hot dogs, franks) and childhood brain cancer (36), and two studies reported associations between frankfurter use and childhood leukemia (37, 38). These are two of the most common childhood cancers (32). Frankfurter consumption was estimated in the pregnant mothers, the fathers, and/or the affected children. The view that NOC were involved seemed plausible because alkylnitrosoureas injected into pregnant rats induce brain tumors and leukemia in their offspring (2, 3). This association was disputed because of the problems of recalling the diet and confounding with other factors and because the nitrite level in frankfurters, but not the incidence of these cancers, has been falling (39). Accordingly, we concentrated initially on the analysis of frankfurters. Salted, dried fish was also examined because this is a risk factor for stomach and nasopharyngeal cancer (35). These risks may be due to food components such as NOC and NOCP (4, 31) and not only to salt. We analyzed sauces because fish sauce has been linked with gastric cancer in China and produced methylnitrosourea (MNU) when it was nitrosated under mild conditions (40, 41) and because many sauces are stored for long periods at room temperature, which might permit NOC and NOCP to accumulate. Finally, cigarette smoke particulates and tobacco were analyzed because they contain nitrosamines that may be involved in the etiology of lung and other types of cancer in smokers and of oral cancer in tobacco chewers (5). We determined NOCP in addition to NOC because this should help in the evaluation of the ability of the products to generate NOC in vivo after the foods are consumed and there are few previous reports on such analyses. Our general approach was to analyze aqueous extracts of the products for (a) NOC by analysis after addition of SA to destroy nitrite and (b) NOCP by nitrosation under mild conditions, followed by the addition of SA and analysis for NOC.
J. Agric. Food Chem., Vol. 49, No. 12, 2001 6069 MATERIALS AND METHODS In view of the carcinogenicity and volatility of many NOC, they were worked with in a chemical hood and all necessary precautions were taken. Unused NOC were destroyed with aluminum/nickel alloy in alkali (42). Yellow lighting was used because NOC are photolabile (43). NaNO3 was obtained from Mallinckrodt (Paris, KY). N-Nitrosoproline (NPRO) was synthesized (44). Other NOC were synthesized from the corresponding amines or amides and NaNO2/HCl and characterized by their ultraviolet and 1H nuclear magnetic resonance spectra. A sample of MNU was obtained from Sigma (St. Louis, MO). We purchased all other chemicals from Aldrich (Milwaukee, WI) and most glassware from Ace Glass (Vineland, NJ). Analytical System for NOC. This was modified from previous procedures (13, 17-19) and consists of an NOC detection system linked to a TEA (model 402, Thermedics, Waltham, MA) and an integrator (Hewlett-Packard, Avondale, PA). The reaction vessel is a 500-mL round-bottom flask with three 24/40 necks and a No. 7 “Ace-thred” neck and is placed in a heating mantle. The 24/40 necks are connected to (a) a glass gas inlet tube (Ace catalog no. 5295-12) extending 3.5 in. below its joint, a regulator and flow meter in the TEA, and an argon cylinder, (b) a 6-in. immersion thermometer fitted with a thermometer adapter, and (c) a 30-cm-long Graham condenser (Kontes, Vineland, NJ) cooled with an immersible pump (Little Giant 1 series, VWR, Batavia, IL), circulating iced water. The Ace-thred neck is fitted with a Teflon-lined septum (changed after each experiment), mounted on a nylon bushing (Ace catalog no. 5029), for injecting samples. The top of the condenser leads via a Claisen adapter to an air inlet stopcock and a gas outlet tube. The temperature can be changed by adjusting the heating mantle temperature. Nalgene 180 clear plastic tubing [o.d. 1/2 in., i.d. 1/4 in.; and o.d. 1/4 in., i.d. 1/8 in.) and Teflon tubing (o.d. 1/8 in. and i.d. 1/16 in.), all from Nalgene (Milwaukee, WI)], fitting one inside the other, were used to connect seven 4 × 30 cm gas washbottles (Kimball Kontes, Vineland, NJ, catalog no. 15060-125) with each other, the reaction vessel, and the TEA. Washbottles 1-4 contain, respectively, 60 mL each of 1.5 N NaOH, 5 N NaOH, 99+% NaOH pellets, and anhydrous granular 99+% Na2SO4. Washbottles 5 (with 20 mL of EtOAc), 6 (with 20 mL of acetone), and 7 (left empty) are kept at -30 °C in a freezer (Cryofridge, Baxter, McGaw Park, IL). The inlet of washbottle 7 is fitted with a fritted disk (Kimball Kontes catalog no. 28220-1251). To minimize leaks, the rotary valve in the TEA is bypassed with a three-way stopcock, which can disconnect the NOC assembly or let in air. Joints are lubricated with Lubriseal (Thomas Scientific, Swedesboro, NJ). The washbottles are cleaned and their contents replaced every second experiment. The gas exit from the TEA is led into a chemical hood. HBr Mode. EtOAc (160 mL) is added to the reaction vessel. Oxygen flow for the TEA (25 mL/min), the TEA ozonator, and the argon flow (40 mL/min at 3 psi) are turned on. The TEA attenuation is set to 128. When the TEA vacuum reaches 0.5 mm, the heating mantle is turned on to give a reflux rate of 1 drop/s and a boiling point of 28 °C for the EtOAc (into which the thermometer dips). Concentrated HCl/glacial HOAc (5:95, 15 mL) and, later, 33% HBr in glacial HOAc (7.5 mL, from Fluka Chemika, Milwaukee, WI) are injected with syringes (filled in a chemical hood), and 15-30 min is allowed for the TEA response to return to baseline after each addition. Test samples (usually 100 µL) are injected with syringes (Hamilton, Reno, NV) when the response drops close to baseline (every 7-10 min) and produce sharp spikes in the integrator response lasting 3-4 min. NPRO standards (0.1 nmol, 100 µL of 1.0 µM NPRO prepared daily from 5 mM stock solution) are injected at the beginning, middle, and end of each run; these produced similar responses for 3-4 h. Yields are based on the standards and expressed as nanomoles per milliliter of solution or nanomoles per gram of sample. Common problems are a raised pressure due to leakage at joints or the septum and blockage, which reduces the measured pressure and raises the reflux temperature.
6070 J. Agric. Food Chem., Vol. 49, No. 12, 2001 HCl Mode. The HCl/HOAc reagent was added as in the HBr mode, but HBr/HOAc was omitted. Although Xu and Reed (19) found a stable response in this mode at a reflux temperature of 30-32 °C, the response to NPRO in our HCl system operated at 30 °C was very broad, took 15-30 min to return to baseline, and rapidly decreased on subsequent NPRO injections, probably because of loss of HCl. Hence we lowered the reflux temperature to 26 °C, used 160 instead of 120 mL of EtOAc, and raised the HCl level from 44 to 51 mM. Response then remained stable for >4 h. Analyses of Foods and Tobacco Products for NOC and NOCP. Tabulated results are generally mean values for the complete analysis of two separate batches of each product, with duplicate analysis of the final extract in each case. Unless specified otherwise, each sample was taken from a different brand (for commercial products) or batch (for noncommercial products). Mixtures were shaken with a Vortex mixer. A Sorvall RC-5B centrifuge (Wilmington, DE) operated at 5 °C was used to sediment solids. Solutions were generally evaporated to 5-10 mL in a Buchi rotary evaporator operated at
